Summary
- Approved by Government Order G.O.(Ms) No. 24 of the IT Department on 18 Sep 2020. The full policy booklet is annexed to the order.
- Uses a six-part TAM-DEF framework to judge AI: Transparency and audit; Accountability and legal issues; Misuse protection; Digital divide and data deficit; Ethics (privacy and data protection, human and environmental values); Fairness and equity.
- Turns that into a DEEP-MAX scorecard with seven parameters: Diversity, Equity and fairness, Ethics, Privacy and data protection, Misuse protection, Audit and transparency, and cross geography and society (X).
- Procurement rules: departments consult TNeGA's Centre of Excellence in Emerging Technologies (CEET) before buying AI, require a satisfactory DEEP-MAX score (binary 1/0 to start, a 0 to 10 scale planned), and re-score at least every 6 months (suggested).
- Plans an ACTS blockchain (AI Certification, Transparency and Scorecard) to store scores and training data certificates; until it exists, TNeGA is to suggest an alternative.
- Also covers AI training for citizens, officials, legislators and judges, plus promotion measures (expert groups, open data, AI sandbox).
- No official repeal or replacement was found as of 2026-10-19. Tamil Nadu's 2026 to 2027 AI activity (AI Mission expansion, a Tamil language model) was seen in secondary news only and does not mention superseding this policy.
Who it applies to
- Any authority or body in Tamil Nadu owned, controlled or aided by the state government.
- Cooperatives, trusts, societies, public sector undertakings and boards controlled by the state, or whose officers it appoints.
- Partnerships and joint ventures of the state government.
- Private companies are reached indirectly: as vendors selling AI to these bodies, they must pass the DEEP-MAX evaluation.
- Not extraterritorial.
Key dates
- TNeGA sends the draft policy to Government (e-mail cited in the G.O.)
- G.O.(Ms) No. 24, IT Department, approves the policy
Full date record
| Date | Event |
|---|---|
| 8 Sep 2020 | TNeGA's e-mail submitting the policy (referenced in the G.O.) |
| 18 Sep 2020 | G.O.(Ms) No. 24 approves the policy |
Penalties and enforcement
- No fines or penalties.
- The state may set and revise acceptability criteria based on DEEP-MAX scores and may reject or recall any AI application that fails them.
- Oversight by a Safe and Ethical AI Monitoring Committee headed by the Chief Secretary, with senior officers (including the law department) and academic AI and policy experts.
- TNeGA implements and monitors, in consultation with the IT Department, which can issue revisions.
What it means for an Indian company
- Selling AI to Tamil Nadu government bodies? Expect a DEEP-MAX evaluation before purchase and periodic re-scoring after go-live. Build the evidence once.
- Pre-deployment scoring on bias, privacy and misuse: SC-AI-015 Pre-deployment testing and validation, SC-AI-017 Fairness and discrimination testing of outputs, SC-AI-029 AI-specific security controls (prompt injection, model abuse).
- Diverse, certified training data: SC-AI-010 Training and input data governance (provenance, quality) and SC-AI-013 Data bias and representativeness checks.
- Audit trail of decisions for disputes and litigation: SC-AI-032 Logging and record keeping and SC-AI-024 Explain AI-assisted decisions.
- Six-monthly re-scoring of self-learning systems: SC-AI-020 Performance monitoring and drift detection.
- Vendor evidence pack for the buyer: SC-AI-007 Third-party and vendor AI due diligence and SC-AI-014 Model documentation (model cards, system cards).
- Not selling to TN government? No direct obligation, but DEEP-MAX is a handy, simple scoring template for internal use. NITI Aayog's Part 2 paper cites it as an example framework (see NITI Aayog Responsible AI for All).
Obligations
Full obligation breakdown coming
We break the most requested policies down into numbered obligations mapped to controls. Leave your details and we’ll tell you when Tamil Nadu Safe and Ethical AI Policy is done.
Glossary
- G.O.: Government Order, an executive instruction from a state government.
- TNeGA: Tamil Nadu e-Governance Agency, which drafted and implements the policy.
- CEET: Centre of Excellence in Emerging Technologies, inside TNeGA, consulted before AI purchases.
- TAM-DEF: the six evaluation dimensions (Transparency and audit, Accountability and legal, Misuse protection, Digital divide and data deficit, Ethics, Fairness and equity).
- DEEP-MAX: the seven-parameter scorecard used to rate an AI system.
- ACTS blockchain: a planned public ledger for DEEP-MAX scores and training data certificates.
- Data deficit: when a community has too little good data, so AI works worse for it.
Key takeaways
- An executive order binding on Tamil Nadu government bodies only; no penalties, no direct private sector duties.
- Core tool: TAM-DEF framework plus the seven-point DEEP-MAX scorecard, applied before roll-out and roughly every six months after.
- Vendors to the TN government should prepare DEEP-MAX evidence up front.
- "Non-binding" is accurate for private companies but understates it for state bodies, which the policy says must strictly adhere.
Official sources
- G.O.(Ms) No. 24, IT (C.1) Department, 18.09.2020, with annexed policy (cms.tn.gov.in PDF), accessed 2026-10-19
- NITI Aayog Responsible AI Part 2 (August 2021), Appendix 1 citing the TN policy, accessed 2026-10-19
- Not reachable on 2026-10-19: the kygcms.tn.gov.in mirror of the G.O. (connection failed) and a tnprivatejobs.tn.gov.in copy of the booklet (404).
This is general information, not legal advice. Last verified 19 Oct 2026 against the official source. Laws change; confirm the current position with qualified counsel before acting.